RAG-Ready with Fabric | Alexandra Savelieva

In the modern world that evolves at an unprecedented pace, enterprise documentation is often outdated, and a lot of important information is in the form of tribal knowledge. Join this session to learn how you can build an adaptive knowledge base from online discussions using hierarchical agglomerative clustering. Microsoft Fabric is used as the platform to build and run a pipeline for continuously turning raw messages from Microsoft Teams into a structured set of crisp insights with LLM-powered data science capabilities. The output of such processing may be efficiently leveraged for building and evaluating Retrieval Augmented Generation (RAG) systems with data drift correction, automatically refreshing FAQ page on a company website, and identifying issues in documentation. Attendees will take away a highly reusable pattern for working with conversational texts with AI at scale.